Can you say no to your doctor using an AI scribe? (theconversation.com)

Australia’s Therapeutic Goods Administration (TGA) has announced that certain AI scribes—digital tools that record, transcribe, and draft clinical notes during doctor-patient consultations—now legally qualify as medical devices. This classification requires these AI systems to undergo regulatory scrutiny for safety, accuracy, and transparency, marking a significant shift in oversight. The move aligns with similar steps taken by UK health authorities and addresses growing concerns about the reliability, privacy, and ethical use of AI in sensitive healthcare settings. AI scribes promise reduced administrative burdens for clinicians and more attentive patient interactions, but challenges remain. Large language model-based scribes can hallucinate or misinterpret information, risking errors that must be caught by busy doctors. Accuracy also varies with accent and background noise, raising safety concerns in multicultural populations. Privacy risks are acute, given varying data storage practices and potential exposure to breaches. Patients are urged to demand transparency about data use, consent procedures, and options to pause or opt out, especially during sensitive consultations. The TGA’s new regulations require clear audit trails and clinician review before AI-generated notes enter medical records. While the TGA’s action is a crucial step toward safer AI integration in healthcare, experts stress the need for coordinated standards on consent, independent performance evaluations, and adaptive risk-based regulations to ensure these tools genuinely enhance care rather than create hidden burdens or compromise patient trust.